Experience

• Set up and clustered two NVIDIA DGX Spark nodes to host and serve models on-prem, including air-gapped workloads.
• Built a GenAI platform end to end on FastAPI, React 19, Neo4j and ChromaDB: hybrid GraphRAG retrieval, answers streamed over SSE, and multimodal OCR pipelines.
• Brought a critical offline platform on RHEL and Docker Swarm back to life: upgraded 15+ microservices, then ran a PostgreSQL disaster recovery, salvaging data from a replica, rebuilding the corrupted primary and finishing the migration.

• NVIDIA DGX Spark
• FastAPI
• React 19
• Neo4j
• ChromaDB
• GraphRAG
• PostgreSQL
• Docker Swarm
• RHEL

Projects from this role: Air-Gapped Document Assistant, Foundry Vision and Traceability, Real-Time Mosquito Tracking, Gift Registry and Marketplace

• Rewrote a genetic algorithm for coal-blend optimization to score the whole population in one vectorized pass, cutting a run from 5 hours to under 2 minutes (150x).
• Built a thermal-video pipeline that measures hot sinter height at the discharge end: an industrial digital twin operators use to tune fuel rates in real time.
• Built and deployed multi-regression models predicting Internal Return Fines (IRF) and Blast Furnace Return Fines (BFRF) from live process data on a PI Server.
• Deployed a naval ship identification system on an NVIDIA Jetson AGX Orin: YOLO finds the ship and OCR reads its hull number, at 92% accuracy in the field.

• Python
• NumPy
• Genetic Algorithms
• OpenCV
• YOLO
• OCR
• Jetson AGX Orin
• PI Server

Projects from this role: Coal-Blend Optimizer, Real-Time Mosquito Tracking

• Designed the AWS IoT backend, on IoT Core, Lambda, RDS and S3, that onboards devices and ships them over-the-air firmware updates.
• Built a Raspberry Pi telematics unit that reads a vehicle's OBD-II port and GPS and reports live over 4G.
• Built a hotspot/coldspot detection system using a U-Net segmentation model with a React and Tailwind CSS front end.
• Wrote a Samsung SmartThings Edge driver with a custom webhook integration.

• AWS IoT
• Raspberry Pi
• Python
• PyTorch
• React
• Tailwind CSS
• PostgreSQL

Projects from this role: AWS IoT Telematics & Vehicle Diagnostics, Thermal Hotspot/Coldspot Detection

• Led a team of three building an emergency dispatch simulator, with LSTM networks forecasting where calls would come from.
• Split the county into each station's nearest-coverage zone with dynamic Voronoi partitioning, to route ambulances and other emergency vehicles.
• Gathered requirements from the county's emergency staff, so the simulation matched how dispatch actually runs.

• Python
• TensorFlow
• LSTM
• Spatial Geometry
• Pandas

Projects from this role: Emergency Dispatch Simulator

• Cut LMS app load times from 6 seconds to milliseconds by parallelizing data fetching.
• Built a full-stack property-management CRM with React, Flutter, Firebase, Django and Twilio.
• Wrote a custom iOS BLE library that doubled command throughput in a React Native robot controller app.
• Led the AWS Amplify integration for an educational app that reached 5,000+ downloads on the Play Store.

• React
• React Native
• Flutter
• Django
• Firebase
• Twilio
• BLE
• iOS
• AWS Amplify

• Added automated linting to CI, fixed the 3,000+ problems it flagged, and made it a gate for new code.
• Fixed Android 12 native build failures in the app's real-time communication libraries.
• Built a custom video player UI and kept playback state saved locally while streaming.
• Cut 20 MB from the release build by removing dead code and unused dependencies.

• Flutter
• Dart
• Android
• CI/CD
• State Management